Mixed Methods Research: Combining Qualitative and Quantitative Tools
Mixed methods research is not about using more tools. It is about designing a study where quantitative and qualitative components each answer a distinct part of the research question — and then integrating the findings in a way that produces insight neither strand could generate alone.
Why mixed methods — and when not to use it
Mixed methods is appropriate when the research question has two genuinely distinct dimensions: one that can be answered with numbers (how much, how many, how often, with what probability) and one that requires interpretation (why, how, what does it mean to those involved). If both dimensions can be answered by either quantitative or qualitative data alone, adding the other strand adds complexity without insight.
The most common misuse of mixed methods is adding a qualitative component to a quantitative study purely to appear comprehensive. Interviews that are not designed to answer a specific research question, and whose findings are not integrated with the quantitative results, are not mixed methods — they are parallel studies reported in the same document.
The three core designs
Parallel collection, integrated analysis
Quantitative and qualitative data are collected concurrently and independently, then compared and merged. Best when you want to validate or triangulate findings. The integration point is the analysis, not the collection.
Tools: SPSS/R + NVivo/ATLAS.ti → integrated matrix
Quant first, qual explains
Quantitative phase runs first and identifies patterns or anomalies that need explanation. Qualitative phase is then designed specifically to explore those findings. The most common design in evaluation research.
Tools: SPSS/Stata → results → purposive sampling → NVivo
Qual first, quant confirms
Qualitative phase explores the phenomenon and generates categories or a model. Quantitative phase tests whether the model holds at scale. Used in scale development, theory building, and under-researched topics.
Tools: NVivo/ATLAS.ti → construct development → SPSS/AMOS
Tool stack for mixed methods research
| Function | Tool options | Notes |
|---|---|---|
| Quantitative analysis | SPSS, R, Stata | Choice depends on method complexity. R is the strongest long-term investment. |
| Qualitative coding | NVivo, ATLAS.ti, MAXQDA | All three are mature. NVivo dominates in social sciences; MAXQDA has strong mixed methods integration features. |
| Survey collection | KoBoToolbox, SurveyCTO, Qualtrics | Qualtrics integrates mixed data types; KoBoToolbox is free for NGOs and research organisations. |
| Interview transcription | Otter.ai, Rev, manual | AI transcription has improved significantly; manual review always required for research-grade transcripts. |
| Integration and joint display | MAXQDA (mixing tools), Excel matrix, R | A joint display table or figure showing both strands side by side is the standard integration artefact. |
| Report production | Word, R Markdown, Quarto | R Markdown can integrate quantitative output and qualitative quotes in a single reproducible document. |
The integration problem — and how to solve it
The hardest part of mixed methods research is not data collection or analysis — it is integration. Most researchers who claim to use mixed methods produce two parallel reports: a quantitative section and a qualitative section, with a brief paragraph at the end saying "the qualitative findings confirmed the quantitative results." This is not integration.
True integration happens when the findings from one strand change how you interpret the other — when a qualitative theme explains a quantitative anomaly, or when a quantitative pattern directs you to specific cases for deeper qualitative exploration.
The test of integration: Could someone read only the quantitative section and draw the same conclusions? If yes, the qualitative component added nothing. The findings should be genuinely interdependent.
Mixed methods research designed to produce real insight.
Claryon designs and delivers mixed methods studies for evaluation, policy research, and institutional learning — with genuine integration built into the design from day one.